The face-name paired-associates task: an fmri protocol that reliably elicits hippocampus activation Jonas Persson 1, Lars-Göran Nilsson 1 & Lars Nyberg 2 1 Department of Psychology, Stockholm University, 106 91 Stockholm, Sweden. 2 Departments of Integrative Medical Biology (Physiology) and Radiation Sciences (Diagnostic Radiology), Umeå University, 90187, Umeå, Sweden Correspondence to: Jonas Persson Department of Psychology, Stockholm University Frescati Hagv. 14 106 91 Stockholm Sweden E-mail: jonas.persson@psychology.su.se Fax: +46 (0)8-159342 Phone: +46 (0)8-163871 1
Abstract We describe a functional magnetic resonance imaging (fmri) protocol that reliably activates the hippocampus (HC) in human participants. The protocol is based on the face-name pairedassociates task (FN-PA) and relies on processes involved in forming associations and binding together objects in memory. The FN-PA task is not language specific and can be completed in less then 10 min which makes it appropriate for children and older adults. The protocol can be implemented on any MR scanner capable of functional imaging, and has proven valuable for HC activation in young adults. Also, the method has potential for the study of HC processing in aging as well as pathological conditions such as schizophrenia, Alzheimer s disease, and depression. 2
Structures within the medial temporal lobe (MTL) are critical for the forming of new associations, and the importance of hippocampal structures in memory formation is well documented e.g. 1, 2. In particular, episodic memories (i.e. memory for past experiences) that are rich in contextual detail rely heavily on HC involvement. Functional neuroimaging, such as functional magnetic resonance imaging (fmri), is a powerful tool for investigating HC involvement in cognitive tasks, and can also provide information about changes related to aging, as well as pathophysiology underlying various behavioural disorders. Although a growing number of fmri studies specifically investigate hippocampal involvement in various cognitive tasks, many of them fail to find activation in the HC 3. This is challenging in several ways, but can be especially problematic when testing for differences between treatment and control groups, or investigating alterations in HC activation with age. It is therefore imperative to construct and describe tasks that reliable elicits HC activation, and can be used to test hypotheses in a variety of contexts and across study populations. In preparation for a large-scale study on aging, we revised and evaluated the face-name paired-associates (FN-PA) task 4, 5 to address specific needs including a short scanning time, ease of administration to clinical populations and across the age span, and its ability to evoke reliable hippocampus activation. Wheras several different tasks have proven useful for examining hippocampus activation in fmri and positron emission tomography (PET) studies, we were particularly interested in a task that robustly shows hippocampus activation during both encoding and retrieval and also provide a measure of memory performance. The FN-PA task can be used to investigate theories of episodic memory function in general and hippocampal function in particular. Also, given the findings of hippocampal abnormality in aging and dementia, the FN-PA task can be a helpful tool in studies on hippocampal activation in these populations. For example, the brains response to medical treatment and cognitive interventions in aging and dementia can be monitored. 3
In addition to a role for the HC in memory impairments with age and dementia, it has been implicated as a key region in several clinical disorders including schizophrenia and depression. For example, the long-term memory impairments associated with the pathophysiology of schizophrenia may be a direct result of hippocampal circuit dysfunction 6. This is supported by findings of neuroanatomical and biochemical abnormalities in the hippocampus in patients with schizophrenia. Likewise, recent investigations suggest that the hippocampus and altered hippocampal neurogenesis have been implicated in depression 7-9. Together, these findings indicate a significant role for the hippocampus in various psychological disorders in which further investigations using fmri may be of great importance. Recent accounts of declarative memory function in humans have emphasized the role of the HC in forming relational long-term memory representations 10. According to this view, the HC participates in the formation of representations linking multiple items. An essential feature of relational binding is that associations are formed between multiple elements, and that the discrete elements and the relationships among them are accessible 11. That is, the distinct parts are not bound together into one rigid inseparable representation. The HC receives perceptual information from associative regions, and serves to represent relationships among distinct percepts. Thus, including an encoding condition that requires binding of elements in memory should increase the likelihood for reliable HC activation. In the current study we used both a within-modality (visual presentation of both the face and name) and between-modality (visual presentation of the face and auditory presentation of the name) versions to assess whether binding between visual and auditory items would elicit stronger HC activation. One instantiation of binding is when experiences are formed by associating a particular item with information about the context in which the item was studied (i.e. source 4
information). The retrieval of source information may involve a distinct psychological process (recollection) compared to retrieving context-free item information based on the sole feeling of familiarity 12, 13, and it has been suggested that only the former process may engage the hippocampus. Hence, the absence of hippocampal activation in previous investigations of episodic memory retrieval may be explained by the use of familiarity-based recognition tasks. For example, Eldridge and colleagues found reliable HC activations only when correct recognition decisions were based on recollection 14. By including a forced-choice cued recall condition with three alternatives, we hoped to increase the dependency on context attributes, and hence recollection at retrieval. Previous investigations suggest that the HC, in contrast to other MTL structures, is especially susceptible to this manipulation. Based on these theoretical considerations and findings from neuroimaging studies 4, 5, suggesting HC involvement in involvement in relational binding processes, we examined whether reliable HC activation could be elicited using a revised version of the FN-PA task. In addition, it was crucial to develop a task that could be used in different settings using a range of participant populations. Therefore, task characteristics should include: (a) reliable activation of the HC, (b) hypothesis-driven support for a mechanism explaining why the task would be expected to recruit the HC, (c) collection of important behavioural data such as RT and accuracy, (d) instructions that are straightforward so that the task can be performed by participants with impaired cognition and by participants across a wide age range, (e) short duration since certain populations have difficulties with extended scanning sessions, (f) language independence to facilitate cross-cultural comparisons. The FN-PA protocol presented here used a blocked design adapted for fmri, single photon emission computed tomography, or PET studies, although it can easily be modified for use with event-related fmri, event-related potentials, magnetoencephalography or offline behavioural studies. The protocol requires no significant modification to standard 5
fmri experimental procedures for data collection or analysis, and represents a straightforward approach for eliciting hippocampal responses associated with episodic memory. Although the protocol was evaluated using young adults, it is suitable also for children and older adults. Depending on the participants age and cognitive status additional pilot testing may be required in order to optimize order and rate of stimulus presentation. Protocol Materials Human subjects: Handedness is not crucial but should be assessed at least by subject selfreport (preferably documented by a handedness inventory, such as the Edinburgh Handedness Inventory). CRITICAL Subjects need to have normal or corrected-to-normal vision. Vision is usually corrected by personal contact lenses or fmri compatible glasses that are provided by the MRI center. Personal glasses can not be used since these may contain metal and may lead to artefacts or signal dropout. CAUTION The study protocol must be approved for use by the appropriate Human Subjects Committee or Institutional Review Board. Informed consent must be obtained following the established institutional or national guidelines. Equipment 1 MRI scanner capable of functional imaging. With the rapid acceptance of fmri in clinical applications, as well as the great interest in fmri as a research tool, most MR scanners are now equipped with functional imaging sequences using rapid echo-planar imaging techniques to measure Blood Oxygen-Level Dependent (BOLD) contrast associated with neuronal activation. Such sequences are highly suitable for this protocol. Functional MRI equipment and scanning techniques can appropriately vary. Our recent fmri studies have been carried out on a Philips 3.0 tesla high-speed echo-planar imaging device using a quadrature headcoil. 6
For the current study the following parameters were used: repetition time: 1512 ms (31 slices acquired), echo time: 30 ms, flip angle: 70 degrees, field of view: 22 22 cm, 64 64 matrix and 4.65 cm slice thickness. To avoid signals arising from progressive saturation, ten dummy scans were performed prior to image acquisition. 2 fmri analysis software. Functional MR analysis packages are freely available from web resources (Statistical Parametric Mapping (SPM) from the Wellcome Department of Imaging Science, Functional Imaging Laboratory http://www.fil.ion.ucl.ac.uk/fil.html; and Functional Software Library (FSL) from the Oxford Centre for Functional Magnetic Resonance of the Brain http://www.fmrib.ox.ac.uk/), as well as from a number of commercial packages. 3 Scanner compatible button-press response device (see equipment setup) 4 Padded scanner couch 5 Foam ear plugs 6 Head stabilizer (e.g. foam padding within a head coil, or a plastic bite bar molded to each subject s dentation) 7 Stimulus generator. A number of commercial systems exist for the delivery of visual and auditory stimuli in a MR environment. Most MR scanners have an audio delivery system using headphones or a loudspeaker for providing instructions to subjects. We have found that a system employing headphones is adequate for this protocol. Given the strong overlap between the within-modality and between-modality versions of the task (see anticipated results) and the fact that a few participants complained of difficulties hearing the names, we strongly recommend the within-modality version of the task. For scanners with active shielding of the static magnetic field, it is also possible to place a commercial data projector in the scanner room which projects visual data onto a screen placed at the head or foot of the subject that is viewed through a mirror system, thus providing a relatively inexpensive 7
method for presentation of visual stimuli which we have found to be satisfactory for this protocol. The current set of stimuli was generated and presented using E-Prime (Psychology Software Tools, Inc., Pittsburgh, USA; www.pstnet.com/eprime), but any suitable stimulus presentation software/hardware combination that can smoothly display stimuli and record responses (and RT to millisecond accuracy) can be used (see equipment setup). Equipment setup fmri equipment For the button-press response device only three buttons are needed, although a four- or fivebutton setup is typical, in which all buttons but three are ignored. Buttons should be spaced and sized such that they approximate a typical keyboard input device (i.e. participants should be able to comfortably place the index and middle finger of the right hand on the keypad. Whether displayed on a computer screen (for offline studies, PET, MEG, ERPs) or if projected for use during fmri, each stimulus item should be easily read by the participants without strain, but should not take up a large portion of the visual field. Stimuli For the encoding condition, each of the face-name stimuli (Fig. 1) consist of a face shown on a black background with a fictional first name printed to the right-hand side of the face, forming a face-name pair. The name can also appear underneath the face. In addition to visual presentation of names, auditory presentation of names is possible. The faces in the current protocol were digital color photographs taken of individuals previously unknown to the subjects. For each face, additional features such as color of clothes were removed from the picture. There were equal numbers of male and female faces. Popular first names were obtained and assigned to each face by the investigators. For the auditory presentation, names were recorded in a female (for female faces) or male (for male faces) voice, and each of the 8
names was synchronized for onset time (i.e. each auditory presentation began at 1000ms after visual presentation of the face). For the retrieval task, each face was presented with three letters of which one letter corresponded to the first letter in the name-face pair (Fig. 1). For both encoding and retrieval conditions, a fixation cross was presented in the center of each stimuli. For the low-level control condition, a fixation cross was presented in the center of the visual field. The fixation cross was randomly replaced with a circle (see procedure). Procedure Subjects 1 Obtain informed consent from subjects and document handedness and vision Psychophysical procedures 2 Instruct the subjects that the experiment consists of three separate segments that are repeated over the course of the scanning session. 3 Instruct the subjects that during the encoding phase, faces and names will be presented by both (i) the face and name appear on the screen or (ii) the face appears on the screen and the corresponding name is presented auditory via the headphones. One name is coupled with one face and the task is to remember the name for each of the faces (Fig. 1). 4 Instruct the subjects that the faces will subsequently be presented together with three letters, and that the task is to indicate the letter that corresponds to the first letter of the name that was presented together with the face (Fig. 1). The top letter corresponds to the index finger, the middle letter to the middle finger, and the bottom letter corresponds to the ring finger. Inform the subjects that if they do not remember a particular face-name pair they can respond by guessing. Other configurations are possible, and modifications in accordance to specific study requirements are unlikely to affect the results. 5 Instruct the subjects that between study and retrieval blocks, a fixation cross will appear. The task is to indicate, as quickly as possible, when the fixation cross changes to a circle. 9
Depending on individual study parameters and goals, subjects can theoretically use either hand to respond. 6 After that instructions are reviewed, and just prior to entering the scanner (or being formally tested) the subjects should be familiarized with the task by completing a 1-min practice version of the task. CRITICAL STEP Review the responses immediately and monitor the subject while performing the task to ensure that the subject completely understands the task and can perform it correctly. Make sure that the subject is using the correct finger to indicate the letter corresponding to the each face. 7 Have subjects lie on a padded scanner couch in a dimly-illuminated room. Give the subjects earplugs and make sure that they know have to use them. This is important for reducing highintensity scanner noise but allow for spoken instructions to be heard as well. 8 Stabilize the subject s head. This can be done by either use foam padding within the head coil or using a plastic bite bar molded to each subject s dentition. 9 Have subjects complete 20-s blocks each of (i) FN-PA encoding (ii) FN-PA retrieval (iii) control task that are alternated during the scanning session. The number of blocks may vary depending on study parameters and number of subjects. Encoding of specific face-name pairs should always precede the retrieval of these same face-name pairs. In the current study each encoding and retrieval stimulus was presented for 4s with a fixed interstimulus interval of 1-s. For the control condition, a fixation cross appear for 1500-2500ms and change into a circle which appears for 500ms. The circle is followed by a second presentation of a fixation cross that is presented for 2000-3000ms, resulting in a total presentation time of 5s. Subjects complete 4 trials during each block and with an 24 block experiment (as in the current study) the scanning session will result in a total of 32 face-name pairs and a total scan time of 8 min including the control blocks. Block order will affect the results (e.g. increasing the time between encoding and retrieval conditions) and this can be optimized depending on study 10
parameters. In the current study the following block order was used (first letter: E=encoding, R=retrieval, B=baseline; block number; second letter: A=auditory name presentation, V=visual name presentation): E1V, B1, E1A, R1V, B2, R1A, E2V, B3, R2V, E2A, B4, R2A, E3V, B5, E3A, B6, R3V, E4V, R3A, R4V, B7, E4A, B8, R4A. Timing: Subject preparation: approximately 20 min Scanning and testing: approximately 10 min Troubleshooting The protocol is essentially a stimulus protocol delivered during functional imaging, therefore the usual technical issues associated with fmri also apply to this protocol. Subjects need to have adequate corrective vision. Regular contact lenses can be used in the scanner. Most prescription spectacles, however, cannot be worn in the scanner due to metal components in the frame. Ways of correcting for reduced visual acuity include the use of scanner compatible glasses or interchangeable corrective lenses held in a plastic frame that can be used to quickly correct subject s vision before entering the scanner. Such glasses or lens sets are readily obtained from optometrists. It is important to monitor and review the practice task results in order to make sure that the subjects understand how to perform the task. It should be noted that the blocked format may not be optimal for all studies, and researchers might consider modifying the FN- PA task for an event-related format (see ref. 15 ). Also, older adults, children, and subjects recruited from patient groups may perform the task more slowly and with less accuracy. For these cases, additional pilot testing might be required to establish optimal timing parameters and order of block presentation for the experimental group in question. Anticipated results Behavioral data 11
The number of correct responses in healthy adults should be expected to be high. The mean accuracy for the 17 healthy adults included in the present study was 71% correct responses (with a base rate of 33%) and the mean reaction time (RT) was 2437ms across the visual and auditory conditions. There were no significant behavioural differences between visual and auditory presentation of names. For the non-correct responses which constituted 29% of all responses, 23% were incorrect responses, and 6% were non-responses. fmri data A slightly different version of the FN-PA task has been validated in previous studies of healthy adults, and episodic memory minus control subtractions can be expected to activate the hippocampus including subregions such as the dentate gyrus 4, 5. Data from the present investigation confirm these findings using the current protocol (Fig. 2). Note that although all retrieval condition were similar, we separate between the condition in which the face/name appeared visually during encoding (retrieval visual) and the condition during which the face was presented visually and the name was presented auditory (retrieval auditory). During both encoding and retrieval versus the control baseline, activation was found in bilateral hippocampus (encoding visual: left peaks at x, y, z = -24, -30, -6, and -32, -20, -18; right peak at 24, -28, -4; encoding auditory: left peak at -18, -28, -10 and right peak at 18, -30, -6; retrieval visual: left peak at -24, -18, -12 and right peak at 20, -30, -4; retrieval auditory: left peak at -24, -30, -4 and right peak at 24, -28, -4). Based on the proposal for a functional dissociation along the hippocampal anterior-posterior axis 16, we also directly compared encoding and retrieval conditions. Using a more liberal threshold (P<0.001 uncorrected) these contrasts showed one reliable difference in the posterior part of the hippocampus. This finding is inconsistent with earlier proposals by Lepage and colleagues 16 that the anterior hippocampus is more engaged during encoding and the posterior hippocampus is more involved in retrieval processes. Instead, observations of 12
predominantly posterior HC activation across encoding and retrieval conditions are compatible with suggestions that the posterior hippocampus is more integral to episodic memory 17. One noteworthy observation was a unique HC activation elicited when a visually presented face was paired with an auditory presented name compared to when both name and face were presented visually. One possibility is that increased HC activation is a result of cross-modal integration during encoding, and that this integration may serve as an additional binding factor. In addition to reliable brain responses in the hippocampus, activation was found in regions of the prefrontal cortex (PFC) and the anterior cingulate cortex. The present results showed encoding-related activation in right and left inferior frontal gyrus (Brodmann Area [BA] 44) the anterior cingulate gyrus (BA 6/32), and the left anterior PFC (BA 9). When retrieval was contrasted with the control condition, activation was observed in bilateral inferior frontal gyrus (BA 44/45, and 47), the anterior cingulate gyrus (BA 32), and the left frontal pole (BA 10). A direct contrast of conditions revealed that specific prefrontal regions were recruited during encoding and retrieval (data not shown), which is consistent with previous results 18, 19 Although the exact roles these structures play in encoding and retrieval of episodic memory remain to be specified, convergent observations suggest that these regions interact with the HC to promote successful memory 20. References 1. Scoville, W.B. & Milner, B. Loss of recent memory after bilateral hippocampal lesions. Journal of Neurology, Neurosurgery, and Psychiatry 20, 11-21 (1957). 2. Milner, B., Squire, L. & Kandel, E. Cognitive neuroscience and the study of memory. Neuron 20, 445-468 (1998). 13
3. Henson, R. A mini-review of fmri studies of human medial temporal lobe activity associated with recognition memory.. The Quarterly Journal of Experimental Psychology 3/4, 340-360 (2005). 4. Zeineh, M.M., Engel, S.A., Thompson, P.M. & Bookheimer, S.Y. Dynamics of the hippocampus during encoding and retrieval of face-name pairs. Science 299, 577-580 (2003). 5. Sperling, R.A., et al. Encoding novel face-name associations: a functional MRI study. Human Brain Mapping 14, 129-139 (2001). 6. Boyer, P., Phillips, J.L., Rousseau, F.L. & Ilivitsky, S. Hippocampal abnormalities and memory deficits: New evidence of a strong pathophysiological link in schizophrenia. Brain Research Reviews 54, 92-112 (2007). 7. Becker, S. & Wojtowicz, J.M. A model of hippocampal neurogenesis in memory and mood disorders. Trends in cogntive science. 11, 70-76 (2006). 8. Airan, R.D., et al. High-speed imaging reveals neurophysiological links to behavior in an animal model of depression. Science 317, 819-823 (2007). 9. Sahay, A. & Hen, R. Adult hippocampal neurogenesis in depression. Nature Neuroscience 10, 1110-1115 (2007). 10. Davachi, L. Item, context and relational episodic encoding in humans. Current Opinion in Neurobiology 16, 693-700 (2006). 11. Cohen, N.J. & Eichenbaum, H.A. Memory, amnesia, and the hippocampal memory system. (MIT Press, Cambridge, MA, 1993). 12. Mandler, G. Recognizing: The judgement of previous occurence. Psychological Review 87, 252-271 (1980). 13. Yonelinas, A.P. The nature of recollection and familiarity. Journal of Memory and Language 46, 441-517 (2002). 14
14. Eldridge, L.L., Knowlton, B.J., Furmanski, C.S., Bookheimer, S.Y. & Engel, S.A. Remembering episodes: A selective role for the hippocampus during retrieval. Nature Neuroscience 3, 1149-1152 (2000). 15. Dale, A.M. & Buckner, R.L. Selective averaging of rapidly presented individual trials using fmri. Human Brain Mapping 5, 329-340 (1997). 16. Lepage, M., Habib, R. & Tulving, E. Hippocampal PET activations of memory encoding and retrieval: The HIPER model. Hippocampus 8, 313-322 (1998). 17. Moser, M.-B. & Moser, E.I. Functional differentiation in the hippocampus. Hippocampus 8, 608-619 (1998). 18. Nyberg, L., Cabeza, R. & Tulving, E. PET studies of encoding and retrieval: The HERA model. Psychonomic Bulletin & Review 3, 135-148 (1996). 19. Habib, R., Nyberg, L. & Tulving, E. Hemispheric asymmetries of memory: the HERA model revisited. Trends in Cognitive Sciences 7, 241-245 (2003). 20. Simons, J.S. & Spiers, H.J. Prefrontal and medial temporal lobe interactions in long-term memory. Nature Reviews Neuroscience 4, 637 648 (2003). Figure captions Fig. 1. Representative samples of stimulus material for the encoding, retrieval and control conditions respectively. Fig. 2. (a) sagittal sections of an anatomical template brain on which are superimposed loci where activation was significantly greater across memory conditions compared to baseline 15
(FEW corrected at p<0.05) (b) schematic rendering of fmri encoding and retrieval activations. Foci from the left hemisphere are shown on a single saggital plane taken from the Talairach and Tournoux (1988) atlas (25 mm lateral to the midline). Peak activations at VISUAL/VISUAL encoding (open square), VISUAL/AUDITORY encoding (open circle), VISUAL/VISUAL retrieval (filled square), and VISUAL/AUDITORY retrieval (filled circle). Fig. 3. Coronal sections of an anatomical template brain on which are superimposed loci where activation was significantly greater during VISUAL/VISUAL encoding (top left) and retrieval (bottom left) trials then control trials (FDR corrected threshold at p<0.05). Middle panel shows magnitude estimates (% signal change) for each of the conditions respectively. In order to assess the robustness of the data across participants, we plotted the magnitude estimates (% signal change) for each individual subject (right hand panel). Note that the majority of subjects show a similar pattern of activation (i.e. encoding and retrieval > control). Additional follow up analyses also showed that there was no relationship between HC activation and performance. Fig. 4 Coronal sections of an anatomical template brain on which are superimposed loci where activation was significantly greater during VISUAL/AUDITORY encoding (top left) and retrieval (bottom left) trials then control trials (FDR corrected threshold at p<0.05). Middle panel shows magnitude estimates (% signal change) for each of the conditions respectively. In order to assess the robustness of the data across participants, we plotted the magnitude estimates (% signal change) for each individual subject (right hand panel). Note that the majority of subjects show a similar pattern of activation (i.e. encoding and retrieval > control). Additional follow up analyses also showed that there was no relationship between HC activation and performance. 16
Encode Recal Baseline